Loudness and dynamics: the loudness race

Popular music got louder from the 1980s to the 2000s, and then streaming stopped it. The story of the loudness race, sometimes called the loudness war, is a good lesson about how a metric can distort a medium.

Why louder wins

Given two versions of the same song, the louder one tends to sound better at first listen. The ear's sensitivity grows with level, see Equal-loudness contours, so a quick comparison favours the louder one. Producers and labels competed to make records louder than the ones on the radio before them.

How it was done

With digital media, the maximum level is fixed at 0 dBFS. To be louder, the average level has to go up while peaks stay below this ceiling. That means shrinking the crest factor (peak minus average level) through:

• heavy compression,
• peak limiting with a high input gain, and
• clipping.

The dynamic range of the music shrinks. A 1980s pop track may have a difference of 14 dB or more between peak and average, while a modern loud master may have 6 dB or less.

The costs

• Distortion from clipping and aggressive limiting.
• Less impact: loud and quiet parts get closer, so crescendos and drum hits lose force.
• Listening fatigue. Constant density tires the ear.
• No gain in the end: if all tracks are loud, none stands out.

Measuring loudness

Peak level says little about perceived loudness. A better measure is LUFS (loudness units relative to full scale), defined in ITU-R BS.1770. It uses a frequency weighting close to the ear's response and gating to ignore silence, and it gives one number for a programme. The loudness range (LRA) describes how much it varies.

Normalisation

Broadcast standards such as EBU R128 specify a programme loudness of −23 LUFS and a maximum true-peak level of −1 dBTP. Many streaming services also normalise tracks to a common target, typically somewhere between −14 and −16 LUFS. If a service turns a loud master down, the extra limiting buys nothing, and the squashed dynamics remain. Many engineers now master to the dynamics they want and leave the rest to the player.

What this means for your work

1. Judge loudness at matched levels, not at the same fader position.
2. Check masters against the target of the distribution channel.
3. Keep peak margins for lossy codecs, which can overshoot by 1 dB or more.
4. Use dynamics as a musical tool: a quiet passage makes the loud one sound loud.

See the level meter scales tool for how digital levels relate to analogue references.

Draft: written by an AI assistant, awaiting technical review.